John's Emails	strip links 8
To: frank@bio.vu.nl Hi Frank, >> IF (inlink1 + inlink2) > 5 >> outlink3 = 10 >> ELSE >> outlink3 = 0 >> ENDIF > > + = plus hup > > = > hup > inlink = I1 I2 > outlink1 = O3 > 10 = 10,0 = 0 > and = and hup > or = or hup > -|\ = links > > I1 I2 > \ / > 5 + > \ | > > > / \ > or and > | | > 0-O3 O3-10 > OK, now I see what you are actually saying. You want to replace the script with a weblet. I hadn't intended to go quite that far yet. It's quite radical. I think we could save a lot of time in future if we use more examples. I think that strictly your diagram should look more like:- I1 I2 \ / + | 5 R \ | > / \ or and | | 0-O3 O3-10 where R = I1 + I2, and R is data (stored in a webaxon in my model) I have a few questions. How does the > hub (I presume you mean hub not hup) know that it is to compare R > 5 or 5 > R? This information is given by the order in my script above, but in your idea I can see no way to distinguish which input is to the left of the > operator and which is to the right. Also, I don't understand the use of or and and hubs, or why you need them. They are not in my script above. O3 should just be one data output but you seem to have it as two in the diagram? How can O3 be a single input to another hub if it is not a single link? > hups nessecary : + > or and and > map data: +(I1,I2) >(5) or O3(0) > and O3(10) > data: I1 I2 > > this is on a (very basic level) when programming we encounter > a lot of cases of > \ / > > > / \ > or and > | | > > we call this a new hup: if> (requires two input) I do like your idea of "micro" weblets. > our map now becomes zoomable: > > I1 I2 > \ / > 5 + > \ | > if> > / \ > 0-O3 O3-10 > >> It will be necessary also to automatically make/delete links. For a new link from >> say existing webneuron A to existing webneuron B, this may best be done using a >> temp websub with two inputs, i.e. >> >> input1 URLA (caller A) >> input2 URLB (target B) >> >> and the temp websub script just something like:- >> >> START >> CREATE_NEURON webaxAB >> Create a new webneuron which will be the webaxon >> of the new link from A to B. Call this new >> webneuron, webaxAB. >> >> MAKE_LINK(URLA, webaxAB) caller to webaxon link >> MAKE_LINK(webaxAB, URLA) webaxon to target link >> END > >> Thus the caller and target are now linked via webaxAB . >> The interpreter would do the work of actually locating and >> adding to the input and output lists in webneurons A and B. > > in my model: > Eedit this.MAP (URLA WEBxAB URLA) If you are not creating a new structure (webaxAB) then where do you store future data for webaxAB? It has got to go somewhere. In the HTML version it could go on the bottom of the webaxAB page, which is a real created structure. By structure I mean hub, webneuron, webaxon, and webstore. In HTML the structures are files. In assembler each structure is a single chunk of RAM. Moving the chunk about in memory doesn't matter since it has a unique 32 bit code "name" to locate it via a physical address look-up table. If the structure is a webaxon then the data is stored at the end of the chunk. Link information is also stored in the chunk. In the case of a webaxon this would just be the 32 bit code "name" of the caller, and the 32 bit code "name" of the target. These code "names" are analogous to URL's in HTML. It would help if you said where "physically" (files, RAM etc) everything goes in your model. >> The initial data in the webaxon would usually be null or undefined. > >> and to delete a link:- > >> START >> DELETE_LINK(URLA, URLB) >> END > > DELETE MAP > or > EDIT MAP (you probably need the last) Why don't you say which link in the map you are deleting? Surely you are not deleting the whole map? > Is clearer now? Yes, basically I think you want to replace the script by a weblet. This is a good idea. I still don't understand the details of your idea. We will need to go over this rigourously so that I can convince myself that it will work in practice. >>>I don't like the idea of applications. >> >> Nor do I, I just use it because it is expected. I will try to stop this. >> there is really only one application - i.e. the whole network. Linking >> applications can be problem for a conventional OS. Best to remove the >> term altogether in relation to weblets. > > In my view what is now considered running an application will be a special > form of drawing maps (you don't see the map) or in other words programming > the computer. When this happens the computer can easily monitor what the user > is doing and on basis of this data optimise the system or suggesting new maps. I have no idea what you are talking about, but it sounds good. >> This sounds most interesting about DNA. Obviously protein formation involves >> a sort of programming language of the DNA. This is something I have not >> followed up. I have concentrated on a larger scale view (neurons and axons) >> and neglected the inner workings of the neuron, except to say that these >> inner workings could possibly be modeled by the webneuron script, >> or attached/nested webfuns. > > Why not turn it around and do it natures way? Please elaborate.